As people impose an increasingly high requirements on performance of a high-end server, more Central Processing Units (CPUs) and Dual In-line Memory Modules (DIMMs) are required on a single node. Because dimensions of a shelf restrict a size of a Printed Circuit Board (PCB) improving hardware density becomes an orientation of efforts in the industry. In an existing method, on a basis of placing a CPU and a DIMM on a traditional single-layer mainboard, a CPU and a DIMM are placed on both the mainboard and a daughter board by combining the mainboard with the daughter board, and then the mainboard and the daughter board are connected by using a connector.
In the prior art, a manner for mounting a daughter board and a mainboard is generally: mounting a stud on the mainboard, and after a connector on the daughter board is adapted to a connector on the mainboard, connecting the daughter board to the stud by using a screw, so as to implement fastening of the mainboard and the daughter board.
However, the manner for mounting a daughter board and a mainboard in the prior art has at least the following defects:
1. Stud strength at each corner tends to be different, which causes imbalance of the daughter board or fragility of mounting and further causes local deformation or even fracture of the daughter board.
2. When the daughter board is being detached, a large force is required for disengaging the connector, which tends to cause damage to the daughter board.
3. An additional tool, such as a screwdriver, is required for detaching the daughter board, it takes a long time to perform a replacement, and screw-free mounting cannot be implemented.